In conclusion, this chapter has shown that a composite post with a sti core and a compli-ant shell provides enhanced adhesion compared to a homogeneous post under pure normal loading and that the pull-o force of the same post can be substantially decreased by the application of shear. Thus, composite posts serve as a basic structure to achieve enhanced and tunable adhesion. Under normal loading, the pull-o force of a composite post with a low h/R is more than nine times higher than the pull-o force of a homogeneous post. The eective adhesion strength of these posts was greater than 1.5 MPa in some cases, which is a high strength for a van der Waals mediated contact. Circular composite posts were found to have up to a 7× reduction in adhesion when a su ciently large shear displacement is applied compared to a 4× reduction for a similar homogeneous post. These composite posts were also used as stamps in a microtransfer printing process.
Chapter 6
Micro-Scale Arrays
Arrays of micro-scale composite posts have the potential for use in a number of broader applications compared to single posts. Arrays can support larger loads as they allow large contact areas. Furthermore, as the load is shared by the array and there is redundancy, the contact of each post becomes less critical. This allows for high adhesion even in dirty environments or if contact on some of the posts is not ideal. Finally, as shown here, arrays of composite posts retain high adhesion when small shear displacements are applied. This is a change from the behavior of individual posts, where shear reduces the eective adhesive strength of the posts, however with su cient shear, the pull-o force of the arrays does decrease.
6.1 Fabrication and Experimental Setup
Arrays of composite posts with a silicon core surrounded by a PDMS layer (Fig. 6.1) were fabricated via a molding process (Fig. 6.2). First the silicon cores were fabricated through a series of micro-fabrication steps. The pattern (Fig. 6.3), both the insets and alignment marks, was directly written with a DWL 66+ (Heidelberg Instruments, Heidelberg, Ger-many) on a Si wafer with a 7 micron thick coating of SPR220-7 photoresist (Dow, Midland, MI). After patterning, the resist was given at least two hours to rehydrate, before being placed on a 115◦C hotplate for two minutes. MF-26A (Dow, Midland, MI) was used to
Figure 6.1: Images of arrays of micro-scale composite posts. Schematic of a single composite post from the side and top, with relevant parameters labeled.
develop the wafer. A deep reactive ion etcher (SPTS, Newport, England) was used to etch the space around the insets to a speci c height which is varied depending on the desired h.
A Remover PG (Microchem, Newton, MA) bath at 60◦C for 10 minutes was used to remove the developed SPR220-7. The molds were made by casting 10:1 PDMS (Sylgard 184, Dow Corning Corporation, Midland, MI) against a silicon master, which is created in a similar manner to the insets. The main dierences are that the radii of the circles in the elements within the written patterns are larger (the radii of the circles was 100 µm rather than 90 µm) and that masters are etched to a depth of 120 µm. The master is treated with O2 plasma and then Trichloro (1H,1H,2H,2H-per uorooctyl) silane (Sigma Aldrich, St. Louis, MO) to facilitate release. The PDMS molds are cured for 48 hours at 25◦C to prevent shrinkage of the features. The molds are also treated with O2 plasma and Trichloro (1H,1H,2H,2H-per uorooctyl) silane. The silicon inset wafer and PDMS mold, lled with uncured PDMS, are aligned on an EVG620 Automated Bond Alignment System (EVG, St. Florian am Inn, Austria). The PDMS is cured for 48 hours at 25◦C under a 10 kg weight. Once the PDMS is cured, the mold is peeled away leaving an array of silicon insets with PDMS shells.
A custom mechanical test system was used to perform pull-o experiments (Fig. 5.3).
A stepper motor-driven translation stage (PRO165, Aerotech, Pittsburgh, PA) and a load cell with a full-scale range of 5 lb (Cooper Instruments and Systems, Warrenton, VA). The array of posts was attached to the load cell atop an x-y translation stage supported on a tip-tilt platform. A at punch indenter, either glass (made from an optical Boro oat
Figure 6.2: Fabrication process ow for arrays of composite posts. Mold fabrication is shown in steps a-r), inset fabrication is shown in steps s-v) and the steps to make composite posts are w-y).
a) and s) Wafers are spun with SPR 220-7. b) and t) The SPR is patterned through a direct write process, developed and then c) and u) etched via DRIE. d) and v) The developed photoresist is removed from the etched wafers. h) A glass slide acts as a spacer for the mold. e) and i) Plasma treating the mold master and the glass slide is followed by f) and j) silanization. g) PDMS is poured into the mold master. k) PET is placed on a silicon wafer and l) then treated with 120 Primer.
m) The glass slides are placed a top the PET and PDMS is poured over this surface. o) The parts made in g) and n) are combined and the PDMS sandwiched between the two wafers is allowed to cure. p) The PDMS mold is taken o the master and then q) plasma treated and r) silanized. The PDMS mold, r), and the silicon inset, v), are w) aligned in a wafer aligner, x) pressed into contact and allowed to cure for 48 hours at 25◦C. y) The composite posts are removed from the mold.
Figure 6.3: Image of pattern for insets. The circle shows the overall outline of a 100 mm wafer. The two small marks on the sides are alignment marks. The arrays of circles are broken up into squares with 12.5 mm side lengths and 10 mm by 50 mm long strips. The patterns in the three squares and two strips have the same size posts and spacing. The diameter of each inset is 190 µm, the center to center distance between pillars is 400 mum, and they are on a hexagonal grid.
window) or PDMS (cast from a glass indenter), with a diameter of 5 mm and 1.75 mm tall was mounted with a perpendicular brace to the translation stage and above the array of posts. A Guppy Pro camera (Allied Vision, Exton, PA) with a lens (NVM-50M23, Navitar, Rochester, New York) or microscope (Precise Eye 1-62840, Navitar, Rochester, NY) tted with a 5× objective is aligned above the indenter to allow for visualization of the contact.
The indenter was rst aligned to the array using the tip-tilt stage by optical observation, and then by tracking the pull-o force and iterating the tip-tilt stage until the pull-o force is maximized. For the normal pull-o force measurements, the post was brought into contact with the punch at a speed of 1 µm/s until a selected preload was reached, the indenter was held at the preload for 30 seconds to ensure the system was stable, and then retracted at a rate of 10 µm/s. When a shear displacement is applied, the pull-o force measurements are performed in a similar manner. The dierences are that after the 30 second hold period a controlled shear displacement was applied followed by a 10 second hold period before the indenter was retracted. The pull-o force in all cases was de ned as the peak tensile force
measured during retraction.
6.2 Normal Loading Results
Initially, normal pull-o force tests using a glass indenter were performed at a range of preloads (Fig. 6.4), to establish what preload was su cient. As seen in Fig. 6.4 for arrays of posts with high h/R ratios, the eect of preload is relatively small; presumably since most posts achieve uniform contact under relatively low preloads. For arrays of posts with lower h/R ratios, the pull-o force increases signi cantly as the preload does, implying that higher preloads are required to ensure all the posts are in conformal contact. The pull-o force of the indenter against a at piece of PDMS is also measured for comparison. The arrays have 23% of the contact area of the at piece of PDMS, yet if the h/R ratio is reduced to 0.21 the same pull-o force as a at piece of PDMS can be achieved.
Figure 6.4: Pull-o force as a function of preload for composite posts with various h/R ratios.
Preload has a minimal eect on posts with high h/R ratios, and large eect on posts with low h/R ratios. The pull-o force of arrays of posts are also compared to that of a at slab of PDMS. Arrays of posts with h/R = 0.21 reach higher pull-o forces than the at slab of PDMS despite signi cantly less area in contact.
As is seen for individual composite posts, arrays with lower h/R ratios have higher adhesion than homogeneous posts. A 3× increase in adhesion for arrays with low h/R ratios compared to homogeneous arrays, was seen with both a glass indenter (Fig. 6.5) and
a PDMS indenter (Fig. 6.6), however the pull-o force against the glass indenter is about 2.5× higher for all arrays than against a PDMS indenter. Also, it is notable that posts with h/R > 0.75 (against a glass indenter) or h/R > 1.0 (against a PDMS indenter) behave similarly to a homogeneous post, suggesting that the inset is too far from the contacting surface to have an eect on the pull-o force.
A 3D LEFM model of an individual cylindrical composite post was used to calculate the strain energy release rate at the interface. 3CD8R elements were used. The elements around the crack tip were 1 µm square in the lateral directions and 3 µm tall. These elements also were organized on a grid aligned with the crack so that VCCT could be used to nd the strain energy release rate as well as the pull-o force. Away from the crack the elements are up to 5 µm in the lateral directions and up to 14.5 µm out of plane. The crack was assumed to be a straight line on one side of the post which was at most 5 µm away from the edge. To scale from the array down to an individual post, it was assumed that all the posts supported an equal part of the load. This FE model predicts G = 0.085 J/m2 (solid line in Fig. 6.5).
One way to compare the pull-o force of arrays and individual posts is to consider the eective adhesion strength. The strength (pull-o force divided by area of the indenter) of the homogeneous arrays is 0.012 MPa. For low h/R arrays, the strength reached 0.04 MPa.
These are signi cantly lower than the strengths associated with individual micro-scale posts (from 0.12 - 1.5 MPa) and individual mm-scale posts (0.07 - 0.2 MPa). However if we only consider the area in contact (23 % of the indenter area) the strengths vary from 0.05 MPa to 0.17 MPa, which are quite similar to the strengths of individual mm-scale homogeneous posts. It is also worth noting that the h/R ratios for the arrays are larger (h/R = 0.21) than that of the individual mm-scale posts (h/R = 0.07), which could explain why the µm-scale array has a similar eective adhesion strength to the mm-scale arrays. If we compare a mm-scale individual post with an h/R ratio of 0.2, it has a strength of 0.14 MPa, which is slightly lower than that of the µm-scale array, since the array has the advantage of contact splitting.
Figure 6.5: Pull-o force against a glass indenter as a function of h/R. Arrays of posts with low h/R ratio have pull-o forces 3× higher than arrays of homogeneous posts. The FE model, solid line, predicts the same behavior as the experimental results, individual points. Each data point is the average of ve measurements and the error bars represent the standard deviation.
Figure 6.6: Pull-o force against a PDMS indenter as a function of h/R. Arrays of posts with low h/R ratio have pull-o forces 3× higher than arrays of homogeneous posts. This trend is similar to that seen with a glass indenter, however the pull-o forces for all h/R ratios are lower. Each data point is the average of six measurements and the error bars represent the standard deviation.
The interfaces were observed optically during the pull-o tests to provide more infor-mation about detachment that purely the pull-o force. The separation mechanism varies depending on whether the post is homogeneous or a composite. For all arrays examined the cracks initiate a the edge of the posts, however the the propagation along the surface re ects the geometry of the posts. As seen in Fig. 6.7 for homogeneous arrays, h/R = 1.2, the crack initiates at an edge of each post and spreads along the circumference of pillar until it reaches a critical point, when it begins to spread across the face of the pillar. While the
Figure 6.7: Optical images showing contact and separation of a select area of an homogeneous array of posts with a glass indenter. a-f) The indenter is moved towards the array. a) The posts and indenter begin out of contact. b) As the indenter approaches, fringes appear suggesting that contact is about to occur. c) The posts and indenter are pressed into conformal contact and d-f) the indenter is pressed further into contact with the array until a preload of 2 N is applied causing a dark gray circle around the post form and then expand. This dark gray circle is from the PDMS bulging resulting in nonuniform diameter viewed from above which causes a discoloration in the image. g-l) The indenter is moved away from the posts. i) Cracks initiate along the edge of the posts (arrows) and then j-k) grow smoothly until l) the posts and indenter are separated.
pull-o force is minimally aected by adding a short inset to the array of composite posts (h/R = 0.95), the mechanism of detachment is quite dierent (Fig. 6.8). For these arrays, as the indenter is retracted a darker circle on the edge of the posts moves towards the center suggesting one of two things. If the PDMS above the inset is well bonded to the inset, then it is contracting inward. If the PDMS and inset are not attached, then it is the thin layer of PDMS around the post getting pulled up above the inset and it is contracting inward. When a crack does develop, it starts at defects along the edge, which is visible as small circles in Fig. 6.8g-h that increase in radius and move towards the posts center. After detachment occurs, the center is the last place to return to the original plane of the surface (6.8i). For the array with the lowest h/R ratio, h/R = 0.21, the separation (Fig. 6.9) is similar to the inset array with h/R = 0.95. During retraction a dark black circle encroaches from the edge towards the center. When a crack initiates it begins at a defect, and then spreads under the inset section of the post (Fig. 6.9g). The crack grows under the inset, since that region has the highest stress concentration, before spreading out and causing the posts to detach.
Under normal loading, arrays of composite inset posts have higher eective adhesion strengths than homogeneous arrays. This enhancement is due to the favorable stress dis-tribution along the contacting interface, similar to that described for single post results as described earlier The crack front for homogeneous posts is a smooth line which moves from one side of the contacted interface to the other. For all composite post arrays of all h/R ratios tested here, the crack initiates from a point defect along the edge. For the high h/R ratios, the crack propagates with an irregular front while for low h/R ratios, the crack spreads to the center of the post, above the inset where the stress is high.